Conversation
The BOOST_UUID_USE_RISCV_V config macro is automatically defined when the compiler defines __riscv_v, indicating RISC-V Vector Extension (V 1.0+) support is available. The macro is also added to the BOOST_UUID_NO_SIMD exclusion list.
This adds a RISC-V Vector (RVV) implementation of the from_chars algorithm, ported from the x86 SIMD implementation. The RVV version supports VLEN >= 128 and handles narrow and wide character types (char, char16_t, char32_t). The implementation follows the same structure as from_chars_x86.hpp: deinterleaving of input character pairs, dash validation, hex character validation, and nibble packing. It uses two separate expected dash vectors (one per character pair) instead of x86's single shuffled vector, as RVV's vrgather-based deinterleaving makes this approach more natural. The generic fallback path is preserved: both the x86 and RISC-V SIMD paths dispatch to from_chars_generic when std::is_constant_evaluated() is true.
Use the number of available characters as the vector length instead of copying them into a zero-initialized temporary buffer. The lanes above n are tail-agnostic, but they are never observed, since the parse has already failed and the result is not stored. With the temporary buffers gone, the memcpy helper is no longer used and its include is dropped.
Template the core on the character type and load each constant where it is used instead of passing seven preloaded vectors, shortening the signature. No change in behavior.
Check __riscv_vector and __riscv_v_min_vlen instead of __riscv_v, so that targets with shorter vectors (Zve32*, Zve64*) use the generic implementation rather than producing incorrect results. The vector length the algorithm requires is now specified where it is used instead of being passed around as a vl parameter.
The valid masks were only negated right after being computed, so a single non-zero test replaces the vmseq/vmnand pair.
All four dashes are at character positions 8..23, so sliding chars1 down by 8 and chars2 into lanes 8..15 gives a single vector to check against one dash constant, instead of two comparisons plus a dash character splat. The middle constant replaces mm_expected_dashes1 and mm_expected_dashes2.
load_vector128 was a one-line wrapper around the load intrinsic. Inline it at its call sites, where the required vector length is already available as the local vl.
e4d08
force-pushed
the
feature/from_chars_rvv
branch
from
September 12, 2026 21:54
b730de9 to
2dafb46
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.